Research And Practice On The SAN Storage Model For Daqing Oilfield

Since 2005, to the actual conditions of Daqing oil field, we fully analyses and study about the data management demand in the oil field, have designed and developed the storage area network system. This text has described the studying and practice about the technology of storage and the administrative of the network data I have done in the course of information construction of Daqing oil field in detail. These technology mainly Include SAN(Storage Area Network), the construction of the storage area network, the design of the model of the tactics of the storage and management , the design of the model of data back up and dispatch,and the design and realization of TSM (Tivoli Storage Management ) storage-backup-monitor and control software system.1, Construction of oil field SAN storage networkSAN stands for Storage Area Network, which is a high-speed memory network independent of network system of the server.In this article, with a view to the whole current situation of data management of Daqing oil field and situation, we try our best to study and realize the management model of data center based on SAN whicn can insure the data safely and efficiently. By saving the data centrally, we dissolve the problems of the transition and storage of massive data. The system model is much more usable, flexible and extendable than the traditional C/S model. It highly improves the safety of data and the efficiency of data management.When designing the SAN frame, we analyze the basic reqirements firstly, and then apply Brocade Fabric switches(2 set ), EMC arrays, IBM3590 tape drive and 3584 tape libaries to the construction of SAN framework. And by the 1000Mbp/s LAN, we connet the 22 subunits to SAN, thus we are ready to save and backup all the 22 subunits'massive data.We design the physical frame of SAN built in the Information Center of the Daqing Oilfield Company which is as shown in the following picture.2,The design of the storage management strategy modelIn order to realize the central storage management of massive data, we have to achieve a model of storage management strategy because once we finish the backup and archive of data, we must know how the data is saved and where it is saved. It means that we must possess a strategy model to manage and schedule the data. In fact, the strategy is metadata that manage the backup and archive data. The strategy includes: policy domain, policy collection, backup group and management class, the material setup policy includes:1. Whether the files fit to be backuped and archived or not?2. How many versions should be kept in the backup server?3. How long should the active and the inactive version should be kept?4. In which part of the memory should the ectypal version be kept?And so on.Based on the understanding of the management strategy and the analysis of the requirement of backup, we classify the Clients on the backup properties, put the clients those have the same or similitude properties into the same policy domain and manage them by the same policy.On this management idea we get the strategy design shown as following:Firstly, we classify the backup data into four types: production data (which always be stored in Oracle database),Office data(which always be stored in Sqlserver database,massive files and ftp or video data. We setup four policy domain correspondingly: Oracle_Domain,Sql_Domain,file_Domain and ftp_Domain, and set the corresponding policy.Thus we finished the design of the frame of data backup.On choosing the mode of data backup we adopted full-backup incremental backup and difference backup unitedly and separately.3,The construction of data backup schedule modelConsidering the massive data and the excessive clients, we decided to adopt the Rate Monotonic model to manage the backup clients, the Rate Monotonic model can be described as following:We define S={t1,t2,…,tn} as a collection that includes n periodical tasks. We describe the tasks as ti=(Ti,Ci,Di,Pi,Ui) (i=1,…,n), Ti denotes the period of tasks occurrence, Ci denotes the longest time that the task cost and Di denotes expire time, that is, ti has to finish in Di, Pi means the priority of the ti, Ui means the ratio of CPU, then Ui=Ci/Ti, int the total task collection, the CPU's ratio is described asBased on the RM model, taking the time and frequency as properties, and taking account of the balance of TSMSever, we give the schedule model as following formulae: S_i: the starting time of task(i) S_i-1:the starting time of task(i-1) C_i-1:the longest backup time of task(i) D_i :the interval of task(i) and task(i-1)(we set it to be 5 to 10 min by experiences)We design the start time schedule table of data backup of information center.4,The Implement of the storage backup management and monitor software system.After the clients complete the backup tasks, the administrator need to know the state of the backup clients, for example, is the backup task executed successfully? Is it carried out as the policy? Only after getting these information can the administrator make the following work plan. To monitor the backup task, we design a software system that fulfills the actual requirement of data backup for Daqing oilfield.On this object, we analyze the background data of TSM(Tivoli Storage Management, a backup software of IBM Company) using the tool named Powerdesigner, and get the function of each table and the relationship among tables, all includes 75 tables. For example, in COLUMS tables the information of system columns is recorded, in TABLES tables the information of system tables is recorded, in ENUMTYPES tables the information of special types is recorded. This three model compose the core of TSM meta data model.On this knowledge, we design and implement the storage backup management and monitor software.Due to the hardwork paid by the project group, we achieve the object that has been setup when the project started: the construction of the Storage Area Network, the design of the clarity management of storage strategy model, the design of schedule model that can be carried out automatically, and the Implement of the storage backup management and monitor software system.